Quick Links
Quick Links
Project Overview
Project Overview
Project Overview
What it is
What it is
What it is
Cairn is a hiking navigation system built around landmarks rather than a moving dot. AR glasses confirm direction while you walk. A phone app carries the planning. Swappable batteries carry an 8–12 hour day. MFA thesis at Purdue — solo, research prototype, never field-deployed.
Hiking navigation runs on a loop: walk, lose confidence, stop, check the phone, walk on. Every loop trades attention on the terrain for attention on a screen — and people stop to check exactly when they are tired and already off-plan.
So the problem was never “display the route better.” It was: treat interruption as part of the navigation problem.
Cairn 是一套围绕地标而不是移动光点构建的徒步导航系统。AR 眼镜在行进中确认方向，手机 App 承载规划，可换电池支撑 8–12 小时的一天。普渡大学 MFA 毕业设计 —— 独立完成，研究原型，未经真实户外部署。
徒步导航是一个循环：走路 → 失去把握 → 停下 → 看手机 → 继续走。每一轮都在拿看地形的注意力换看屏幕的注意力 —— 而人们停下来核对的时刻，恰恰是已经累了、已经偏离计划的时刻。
所以问题从来不是「把路线显示得更好」，而是：把打断本身当成导航问题的一部分。
Cairn is a hiking navigation system built around landmarks rather than a moving dot. AR glasses confirm direction while you walk. A phone app carries the planning. Swappable batteries carry an 8–12 hour day. MFA thesis at Purdue — solo, research prototype, never field-deployed.
Hiking navigation runs on a loop: walk, lose confidence, stop, check the phone, walk on. Every loop trades attention on the terrain for attention on a screen — and people stop to check exactly when they are tired and already off-plan.
So the problem was never “display the route better.” It was: treat interruption as part of the navigation problem.
Cairn 是一套围绕地标而不是移动光点构建的徒步导航系统。AR 眼镜在行进中确认方向，手机 App 承载规划，可换电池支撑 8–12 小时的一天。普渡大学 MFA 毕业设计 —— 独立完成，研究原型，未经真实户外部署。
徒步导航是一个循环：走路 → 失去把握 → 停下 → 看手机 → 继续走。每一轮都在拿看地形的注意力换看屏幕的注意力 —— 而人们停下来核对的时刻，恰恰是已经累了、已经偏离计划的时刻。
所以问题从来不是「把路线显示得更好」，而是：把打断本身当成导航问题的一部分。
Cairn is a hiking navigation system built around landmarks rather than a moving dot. AR glasses confirm direction while you walk. A phone app carries the planning. Swappable batteries carry an 8–12 hour day. MFA thesis at Purdue — solo, research prototype, never field-deployed.
Hiking navigation runs on a loop: walk, lose confidence, stop, check the phone, walk on. Every loop trades attention on the terrain for attention on a screen — and people stop to check exactly when they are tired and already off-plan.
So the problem was never “display the route better.” It was: treat interruption as part of the navigation problem.
Cairn 是一套围绕地标而不是移动光点构建的徒步导航系统。AR 眼镜在行进中确认方向，手机 App 承载规划，可换电池支撑 8–12 小时的一天。普渡大学 MFA 毕业设计 —— 独立完成，研究原型，未经真实户外部署。
徒步导航是一个循环：走路 → 失去把握 → 停下 → 看手机 → 继续走。每一轮都在拿看地形的注意力换看屏幕的注意力 —— 而人们停下来核对的时刻，恰恰是已经累了、已经偏离计划的时刻。
所以问题从来不是「把路线显示得更好」，而是：把打断本身当成导航问题的一部分。
What I made
What I made
What I made
AR interface — a default state showing one landmark and a compass, a zoned field of view, and designed states for landmark-lost, off-route and weather.
Mobile app — route planning, colour-graded trail conditions, offline sync and a pre-departure readiness check.
Wearable hardware — mid-temple swappable batteries, a shoulder-strap charging dock and a 1:1 physical model.
Evidence — two surveys (n=97, n=132), six interviews, a six-person ergonomic study and a seven-person evaluation across all three prototypes.
AR 界面 —— 只显示一个地标和一条罗盘的默认状态、分区的视野规则，以及地标丢失 / 偏航 / 天气三种被设计过的降级状态。
手机 App —— 路线规划、颜色分级的步道状况、离线同步与出发前准备检查。
可穿戴硬件 —— 镜腿中段可换电池、肩带充电坞，以及 1:1 实体模型。
证据 —— 两轮问卷（n=97、n=132）、6 次访谈、6 人人因实验，以及覆盖三个原型的 7 人评估。
How the system splits
How the system splits
How the system splits
Interaction Design
Interaction Design
Interaction Design
The default view — and what earns its way in
The default view — and what earns its way in
The default view — and what earns its way in
The default in-trail view carries one target landmark and a compass strip. Nothing else. Everything else has to earn its way in — by user request, or by rising risk.
The field of view is a constraint, not a canvas. Content is zoned at ±20° / ±30° / ±45°; the center is reserved for immediate action and risk and is never held for long; dense information stays clear of footpaths, rest spots and forks in view.
Landmark guidance carries a correctness state, not just a location. The compass shows a correct-direction range with two variants — heading right, heading wrong — because an outline around a rock tells you where to look, never whether you are right.
行进中的默认视图只有一个目标地标和一条罗盘。没有别的。其余所有信息都必须自己挣得出现的资格 —— 要么用户主动要，要么风险升高。
视野是约束，不是画布。内容分区在 ±20° / ±30° / ±45°；中心留给即时行动与风险，且不长时间占用；密集信息避开视野里的步道、休息点和岔路。
地标指引带的是正确性状态，不只是位置。罗盘显示正确方向范围，并分「方向正确」和「方向错误」两个变体 —— 因为在一块石头上描个边只告诉你往哪看，从不告诉你走得对不对。
The default in-trail view carries one target landmark and a compass strip. Nothing else. Everything else has to earn its way in — by user request, or by rising risk.
The field of view is a constraint, not a canvas. Content is zoned at ±20° / ±30° / ±45°; the center is reserved for immediate action and risk and is never held for long; dense information stays clear of footpaths, rest spots and forks in view.
Landmark guidance carries a correctness state, not just a location. The compass shows a correct-direction range with two variants — heading right, heading wrong — because an outline around a rock tells you where to look, never whether you are right.
行进中的默认视图只有一个目标地标和一条罗盘。没有别的。其余所有信息都必须自己挣得出现的资格 —— 要么用户主动要，要么风险升高。
视野是约束，不是画布。内容分区在 ±20° / ±30° / ±45°；中心留给即时行动与风险，且不长时间占用；密集信息避开视野里的步道、休息点和岔路。
地标指引带的是正确性状态，不只是位置。罗盘显示正确方向范围，并分「方向正确」和「方向错误」两个变体 —— 因为在一块石头上描个边只告诉你往哪看，从不告诉你走得对不对。
The default in-trail view carries one target landmark and a compass strip. Nothing else. Everything else has to earn its way in — by user request, or by rising risk.
The field of view is a constraint, not a canvas. Content is zoned at ±20° / ±30° / ±45°; the center is reserved for immediate action and risk and is never held for long; dense information stays clear of footpaths, rest spots and forks in view.
Landmark guidance carries a correctness state, not just a location. The compass shows a correct-direction range with two variants — heading right, heading wrong — because an outline around a rock tells you where to look, never whether you are right.
行进中的默认视图只有一个目标地标和一条罗盘。没有别的。其余所有信息都必须自己挣得出现的资格 —— 要么用户主动要，要么风险升高。
视野是约束，不是画布。内容分区在 ±20° / ±30° / ±45°；中心留给即时行动与风险，且不长时间占用；密集信息避开视野里的步道、休息点和岔路。
地标指引带的是正确性状态，不只是位置。罗盘显示正确方向范围，并分「方向正确」和「方向错误」两个变体 —— 因为在一块石头上描个边只告诉你往哪看，从不告诉你走得对不对。
Default in-trail view
Default in-trail view
Default in-trail view
Landmark confirmed — heading correct
Landmark confirmed — heading correct
Landmark confirmed — heading correct
Landmark confirmed — heading wrong
Landmark confirmed — heading wrong
Landmark confirmed — heading wrong
Direction Range on the compass
Direction Range on the compass
Direction Range on the compass
Switch Landmark — why this one
Switch Landmark — why this one
Switch Landmark — why this one
Zoning — ±20° / ±30° / ±45°
Zoning — ±20° / ±30° / ±45°
Zoning — ±20° / ±30° / ±45°
Look-Up Pulse — build the habit, don’t replace it
Look-Up Pulse — build the habit, don’t replace it
Look-Up Pulse — build the habit, don’t replace it
The uncomfortable part — they asked for the opposite
The uncomfortable part — they asked for the opposite
The uncomfortable part — they asked for the opposite
Mobile app — planning before you walk
Mobile app — planning before you walk
Mobile app — planning before you walk
Everything that needs thinking lives here: route exploration, complex planning, offline sync and a pre-departure readiness check — used before you leave and at rest stops, never while walking.
Trail conditions are colour-graded green / yellow / red, so risk is scannable before departure instead of discovered on the trail. This surface carries the density so the glasses don’t have to.
所有需要动脑的事都放在这里：路线探索、复杂规划、离线同步和出发前的准备检查 —— 用在出发前和休息点，绝不用在行进中。
步道状况用绿 / 黄 / 红分级，让风险在出发前就能扫完，而不是在路上才发现。这个界面承担信息密度，眼镜才不必承担。
Everything that needs thinking lives here: route exploration, complex planning, offline sync and a pre-departure readiness check — used before you leave and at rest stops, never while walking.
Trail conditions are colour-graded green / yellow / red, so risk is scannable before departure instead of discovered on the trail. This surface carries the density so the glasses don’t have to.
所有需要动脑的事都放在这里：路线探索、复杂规划、离线同步和出发前的准备检查 —— 用在出发前和休息点，绝不用在行进中。
步道状况用绿 / 黄 / 红分级，让风险在出发前就能扫完，而不是在路上才发现。这个界面承担信息密度，眼镜才不必承担。
Everything that needs thinking lives here: route exploration, complex planning, offline sync and a pre-departure readiness check — used before you leave and at rest stops, never while walking.
Trail conditions are colour-graded green / yellow / red, so risk is scannable before departure instead of discovered on the trail. This surface carries the density so the glasses don’t have to.
所有需要动脑的事都放在这里：路线探索、复杂规划、离线同步和出发前的准备检查 —— 用在出发前和休息点，绝不用在行进中。
步道状况用绿 / 黄 / 红分级，让风险在出发前就能扫完，而不是在路上才发现。这个界面承担信息密度，眼镜才不必承担。
How it degrades
How it degrades
How it degrades
Landmark not visible
Landmark not visible
Landmark not visible
The system does not push the user to keep hunting the same target. It falls back to surrounding vegetation and terrain cues and adds a direction arrow to the bottom compass — deliberately off-center, so a hiker who wants to work it out is not overridden.
系统不会逼用户继续找同一个目标。它退回到周边植被和地形线索，并在底部罗盘上加一个方向箭头 —— 刻意不放在中心，这样想自己判断的人不会被覆盖。
The system does not push the user to keep hunting the same target. It falls back to surrounding vegetation and terrain cues and adds a direction arrow to the bottom compass — deliberately off-center, so a hiker who wants to work it out is not overridden.
系统不会逼用户继续找同一个目标。它退回到周边植被和地形线索，并在底部罗盘上加一个方向箭头 —— 刻意不放在中心，这样想自己判断的人不会被覆盖。
The system does not push the user to keep hunting the same target. It falls back to surrounding vegetation and terrain cues and adds a direction arrow to the bottom compass — deliberately off-center, so a hiker who wants to work it out is not overridden.
系统不会逼用户继续找同一个目标。它退回到周边植被和地形线索，并在底部罗盘上加一个方向箭头 —— 刻意不放在中心，这样想自己判断的人不会被覆盖。
Off route
Off route
Off route
A red halo at the edge of view comes first. Only if the user does not correct does a dialog appear, offering snooze, a new trail, or backtracking.
先在视野边缘出现红色光晕。只有用户没有修正，才升级成对话框，提供稍后提醒、更换路线或原路返回。
A red halo at the edge of view comes first. Only if the user does not correct does a dialog appear, offering snooze, a new trail, or backtracking.
先在视野边缘出现红色光晕。只有用户没有修正，才升级成对话框，提供稍后提醒、更换路线或原路返回。
A red halo at the edge of view comes first. Only if the user does not correct does a dialog appear, offering snooze, a new trail, or backtracking.
先在视野边缘出现红色光晕。只有用户没有修正，才升级成对话框，提供稍后提醒、更换路线或原路返回。
Weather
Weather
Weather
A lightweight prompt at the top, re-raised once after an interval if ignored, escalating to a layered weather map only when conditions justify it. The view returns to the trail automatically when weather improves.
顶部一个轻提示，被忽略后间隔一段时间再提一次，只有条件足够严重才升级为分层天气图。天气转好时自动回到步道视图。
A lightweight prompt at the top, re-raised once after an interval if ignored, escalating to a layered weather map only when conditions justify it. The view returns to the trail automatically when weather improves.
顶部一个轻提示，被忽略后间隔一段时间再提一次，只有条件足够严重才升级为分层天气图。天气转好时自动回到步道视图。
A lightweight prompt at the top, re-raised once after an interval if ignored, escalating to a layered weather map only when conditions justify it. The view returns to the trail automatically when weather improves.
顶部一个轻提示，被忽略后间隔一段时间再提一次，只有条件足够严重才升级为分层天气图。天气转好时自动回到步道视图。
Alerts — where the line sits
Alerts — where the line sits
Alerts — where the line sits
The escalation rule came from a participant, not from me: save the ringing, buzzing and sirens for dangerous terrain or an incoming storm. Everything else should be quiet enough to ignore.
这条升级规则来自参与者，不是我：铃声、震动和警报留给危险地形或来袭的风暴，其余的都应该安静到可以忽略。
The escalation rule came from a participant, not from me: save the ringing, buzzing and sirens for dangerous terrain or an incoming storm. Everything else should be quiet enough to ignore.
这条升级规则来自参与者，不是我：铃声、震动和警报留给危险地形或来袭的风暴，其余的都应该安静到可以忽略。
The escalation rule came from a participant, not from me: save the ringing, buzzing and sirens for dangerous terrain or an incoming storm. Everything else should be quiet enough to ignore.
这条升级规则来自参与者，不是我：铃声、震动和警报留给危险地形或来袭的风暴，其余的都应该安静到可以忽略。
Industrial Design
Industrial Design
Industrial Design
Power is a field-operations problem
Power is a field-operations problem
Power is a field-operations problem
Meta Ray-Ban carries 167 mAh in the glasses and 2,940 mAh in the case. Sport sunglasses weigh 20–40 g; smart glasses weigh 40–100 g, and battery plus PCB is 20–30 g of that gap. The energy cannot live on the face.
So the strategy inverted: from charging the glasses once to recharging while moving. In hiking, power is not a desktop charging problem — it is a field operation that has to work while walking, one-handed, with gloves on.
Meta Ray-Ban 眼镜本体 167 mAh，充电盒 2,940 mAh。运动墨镜 20–40 g，智能眼镜 40–100 g，其中电池加 PCB 占 20–30 g。能量不可能都放在脸上。
于是策略反转：从「把眼镜充满一次」变成「在移动中续电」。徒步场景里，供电不是桌面充电问题 —— 它是一次野外操作，必须能边走边做、单手完成、戴着手套也能做。
Meta Ray-Ban carries 167 mAh in the glasses and 2,940 mAh in the case. Sport sunglasses weigh 20–40 g; smart glasses weigh 40–100 g, and battery plus PCB is 20–30 g of that gap. The energy cannot live on the face.
So the strategy inverted: from charging the glasses once to recharging while moving. In hiking, power is not a desktop charging problem — it is a field operation that has to work while walking, one-handed, with gloves on.
Meta Ray-Ban 眼镜本体 167 mAh，充电盒 2,940 mAh。运动墨镜 20–40 g，智能眼镜 40–100 g，其中电池加 PCB 占 20–30 g。能量不可能都放在脸上。
于是策略反转：从「把眼镜充满一次」变成「在移动中续电」。徒步场景里，供电不是桌面充电问题 —— 它是一次野外操作，必须能边走边做、单手完成、戴着手套也能做。
Meta Ray-Ban carries 167 mAh in the glasses and 2,940 mAh in the case. Sport sunglasses weigh 20–40 g; smart glasses weigh 40–100 g, and battery plus PCB is 20–30 g of that gap. The energy cannot live on the face.
So the strategy inverted: from charging the glasses once to recharging while moving. In hiking, power is not a desktop charging problem — it is a field operation that has to work while walking, one-handed, with gloves on.
Meta Ray-Ban 眼镜本体 167 mAh，充电盒 2,940 mAh。运动墨镜 20–40 g，智能眼镜 40–100 g，其中电池加 PCB 占 20–30 g。能量不可能都放在脸上。
于是策略反转：从「把眼镜充满一次」变成「在移动中续电」。徒步场景里，供电不是桌面充电问题 —— 它是一次野外操作，必须能边走边做、单手完成、戴着手套也能做。
Where the battery sits, and how it swaps
Where the battery sits, and how it swaps
Where the battery sits, and how it swaps
Six participants, four positions, 20 g simulated masses, a −5 to +5 comfort scale — tested through head-up, head-down, jumping and sprinting. Mid-temple won at +1.7. Lower temple +0.5. Behind the ear −0.2, swaying and conflicting with hats and hoods. Front frame −2.3, pressing the nose bridge and competing with the optics.
Four attachment concepts were judged against one-handed alignment, stability, quick removal, fit and glove operation. Magnetic alignment with a snap-fit latch and pogo pins won, because the action reduces to three steps: align, insert, lock. A vertical slide drops the battery if the latch doesn’t seat; a whole-temple slide is blocked by hats; a knob needs two hands.
A shoulder-strap dock holds and charges the spares, moving part of the load off the head. A hole at each temple end lets the glasses hang on the chest during short stops — the case the main flow had missed.
6 名参与者、4 个位置、20 g 模拟配重、−5 到 +5 的舒适度量表 —— 经过抬头、低头、跳跃和短跑测试。镜腿中段以 +1.7 胜出。镜腿下方 +0.5。耳后 −0.2，会晃，且和帽子、连帽衫冲突。镜框前部 −2.3，压鼻梁，还和光学部件抢位置。
4 个固定方案对照单手对准、稳定性、快速取下、适配与戴手套可操作五条标准。磁吸对准 + 卡扣 + 弹簧针胜出，因为动作可以简化成三步：对准、插入、锁定。竖向滑入在卡扣没坐实时会掉电池；整条镜腿滑入被帽子挡住；旋钮式需要两只手。
肩带充电坞收纳并给备用电池充电，把一部分负担从头上移走。每条镜腿末端的小孔让眼镜在短暂停留时挂在胸前 —— 那正是主流程漏掉的场景。
Six participants, four positions, 20 g simulated masses, a −5 to +5 comfort scale — tested through head-up, head-down, jumping and sprinting. Mid-temple won at +1.7. Lower temple +0.5. Behind the ear −0.2, swaying and conflicting with hats and hoods. Front frame −2.3, pressing the nose bridge and competing with the optics.
Four attachment concepts were judged against one-handed alignment, stability, quick removal, fit and glove operation. Magnetic alignment with a snap-fit latch and pogo pins won, because the action reduces to three steps: align, insert, lock. A vertical slide drops the battery if the latch doesn’t seat; a whole-temple slide is blocked by hats; a knob needs two hands.
A shoulder-strap dock holds and charges the spares, moving part of the load off the head. A hole at each temple end lets the glasses hang on the chest during short stops — the case the main flow had missed.
6 名参与者、4 个位置、20 g 模拟配重、−5 到 +5 的舒适度量表 —— 经过抬头、低头、跳跃和短跑测试。镜腿中段以 +1.7 胜出。镜腿下方 +0.5。耳后 −0.2，会晃，且和帽子、连帽衫冲突。镜框前部 −2.3，压鼻梁，还和光学部件抢位置。
4 个固定方案对照单手对准、稳定性、快速取下、适配与戴手套可操作五条标准。磁吸对准 + 卡扣 + 弹簧针胜出，因为动作可以简化成三步：对准、插入、锁定。竖向滑入在卡扣没坐实时会掉电池；整条镜腿滑入被帽子挡住；旋钮式需要两只手。
肩带充电坞收纳并给备用电池充电，把一部分负担从头上移走。每条镜腿末端的小孔让眼镜在短暂停留时挂在胸前 —— 那正是主流程漏掉的场景。
Six participants, four positions, 20 g simulated masses, a −5 to +5 comfort scale — tested through head-up, head-down, jumping and sprinting. Mid-temple won at +1.7. Lower temple +0.5. Behind the ear −0.2, swaying and conflicting with hats and hoods. Front frame −2.3, pressing the nose bridge and competing with the optics.
Four attachment concepts were judged against one-handed alignment, stability, quick removal, fit and glove operation. Magnetic alignment with a snap-fit latch and pogo pins won, because the action reduces to three steps: align, insert, lock. A vertical slide drops the battery if the latch doesn’t seat; a whole-temple slide is blocked by hats; a knob needs two hands.
A shoulder-strap dock holds and charges the spares, moving part of the load off the head. A hole at each temple end lets the glasses hang on the chest during short stops — the case the main flow had missed.
6 名参与者、4 个位置、20 g 模拟配重、−5 到 +5 的舒适度量表 —— 经过抬头、低头、跳跃和短跑测试。镜腿中段以 +1.7 胜出。镜腿下方 +0.5。耳后 −0.2，会晃，且和帽子、连帽衫冲突。镜框前部 −2.3，压鼻梁，还和光学部件抢位置。
4 个固定方案对照单手对准、稳定性、快速取下、适配与戴手套可操作五条标准。磁吸对准 + 卡扣 + 弹簧针胜出，因为动作可以简化成三步：对准、插入、锁定。竖向滑入在卡扣没坐实时会掉电池；整条镜腿滑入被帽子挡住；旋钮式需要两只手。
肩带充电坞收纳并给备用电池充电，把一部分负担从头上移走。每条镜腿末端的小孔让眼镜在短暂停留时挂在胸前 —— 那正是主流程漏掉的场景。
Prototype — 1:1 model
Prototype — 1:1 model
Prototype — 1:1 model
Evaluation & Limits
Evaluation & Limits
Evaluation & Limits
What testing changed
What testing changed
What testing changed
Four of six interface changes shared one root cause: the design said where something was, never whether you were right.
“Bearing” was ambiguous, so it became “Direction Range” and shows a range. Landmark highlighting gained a correctness state. Switch Landmark lost the type field and the distance value — testing showed distance doesn’t help a direction judgement — and used the space to say why the system recommends this landmark.
Two of the four changes removed information rather than adding it. The hardware followed the same logic: a decorative block on the wristband read as a screen, so it came off. That is what testing taught me.
六项界面改动里有四项指向同一个根因：设计说了东西在哪，从没说你走得对不对。
「Bearing」有歧义，于是改成「Direction Range」，显示范围。地标高亮加上了正确性状态。Switch Landmark 删掉了类型字段和距离数值 —— 测试显示距离对方向判断没有帮助 —— 把空间让给「系统为什么推荐这个地标」。
四项改动里有两项是删信息，不是加信息。硬件也是同一逻辑：腕带上一块装饰性凸起被读成屏幕，于是拿掉。这是测试教给我的东西。
Four of six interface changes shared one root cause: the design said where something was, never whether you were right.
“Bearing” was ambiguous, so it became “Direction Range” and shows a range. Landmark highlighting gained a correctness state. Switch Landmark lost the type field and the distance value — testing showed distance doesn’t help a direction judgement — and used the space to say why the system recommends this landmark.
Two of the four changes removed information rather than adding it. The hardware followed the same logic: a decorative block on the wristband read as a screen, so it came off. That is what testing taught me.
六项界面改动里有四项指向同一个根因：设计说了东西在哪，从没说你走得对不对。
「Bearing」有歧义，于是改成「Direction Range」，显示范围。地标高亮加上了正确性状态。Switch Landmark 删掉了类型字段和距离数值 —— 测试显示距离对方向判断没有帮助 —— 把空间让给「系统为什么推荐这个地标」。
四项改动里有两项是删信息，不是加信息。硬件也是同一逻辑：腕带上一块装饰性凸起被读成屏幕，于是拿掉。这是测试教给我的东西。
Four of six interface changes shared one root cause: the design said where something was, never whether you were right.
“Bearing” was ambiguous, so it became “Direction Range” and shows a range. Landmark highlighting gained a correctness state. Switch Landmark lost the type field and the distance value — testing showed distance doesn’t help a direction judgement — and used the space to say why the system recommends this landmark.
Two of the four changes removed information rather than adding it. The hardware followed the same logic: a decorative block on the wristband read as a screen, so it came off. That is what testing taught me.
六项界面改动里有四项指向同一个根因：设计说了东西在哪，从没说你走得对不对。
「Bearing」有歧义，于是改成「Direction Range」，显示范围。地标高亮加上了正确性状态。Switch Landmark 删掉了类型字段和距离数值 —— 测试显示距离对方向判断没有帮助 —— 把空间让给「系统为什么推荐这个地标」。
四项改动里有两项是删信息，不是加信息。硬件也是同一逻辑：腕带上一块装饰性凸起被读成屏幕，于是拿掉。这是测试教给我的东西。
What I couldn’t prove
What I couldn’t prove
What I couldn’t prove
The AR prototype was a flat Figma frame shown full-screen in the Quest 3 browser, not a spatial build — so the zoning the design argues for was never tested under AR. It is a reasoned rule, not a validated one.
There was no baseline. I never compared Cairn against a phone, so I cannot claim it reduces interruption — that is the intent, not a result. No task success or time-on-task data was collected. Seven participants, all designers, indoors: they can tell me whether an interface reads, not whether it works on a mountain.
What I would test first: whether EMG gestures hold up in cold, with gloves, holding trekking poles — with real hikers, outdoors, against a phone baseline.
AR 原型是在 Quest 3 浏览器里全屏显示的一个平面 Figma frame，不是空间构建 —— 所以设计主张的分区规则从未在 AR 条件下被测试过。它是一条有依据的规则，不是被验证过的规则。
研究没有基线。我从未把 Cairn 和手机做对照，所以我不能声称它降低了打断 —— 那是意图，不是结果。没有采集任务成功率或任务时长。7 名参与者全是设计背景、在室内完成：他们能告诉我界面读不读得通，不能告诉我它在山上好不好用。
我会先测什么：EMG 手势在低温、戴手套、握登山杖时是否可靠 —— 用真实徒步者、在户外、以手机为基线。
The AR prototype was a flat Figma frame shown full-screen in the Quest 3 browser, not a spatial build — so the zoning the design argues for was never tested under AR. It is a reasoned rule, not a validated one.
There was no baseline. I never compared Cairn against a phone, so I cannot claim it reduces interruption — that is the intent, not a result. No task success or time-on-task data was collected. Seven participants, all designers, indoors: they can tell me whether an interface reads, not whether it works on a mountain.
What I would test first: whether EMG gestures hold up in cold, with gloves, holding trekking poles — with real hikers, outdoors, against a phone baseline.
AR 原型是在 Quest 3 浏览器里全屏显示的一个平面 Figma frame，不是空间构建 —— 所以设计主张的分区规则从未在 AR 条件下被测试过。它是一条有依据的规则，不是被验证过的规则。
研究没有基线。我从未把 Cairn 和手机做对照，所以我不能声称它降低了打断 —— 那是意图，不是结果。没有采集任务成功率或任务时长。7 名参与者全是设计背景、在室内完成：他们能告诉我界面读不读得通，不能告诉我它在山上好不好用。
我会先测什么：EMG 手势在低温、戴手套、握登山杖时是否可靠 —— 用真实徒步者、在户外、以手机为基线。
The AR prototype was a flat Figma frame shown full-screen in the Quest 3 browser, not a spatial build — so the zoning the design argues for was never tested under AR. It is a reasoned rule, not a validated one.
There was no baseline. I never compared Cairn against a phone, so I cannot claim it reduces interruption — that is the intent, not a result. No task success or time-on-task data was collected. Seven participants, all designers, indoors: they can tell me whether an interface reads, not whether it works on a mountain.
What I would test first: whether EMG gestures hold up in cold, with gloves, holding trekking poles — with real hikers, outdoors, against a phone baseline.
AR 原型是在 Quest 3 浏览器里全屏显示的一个平面 Figma frame，不是空间构建 —— 所以设计主张的分区规则从未在 AR 条件下被测试过。它是一条有依据的规则，不是被验证过的规则。
研究没有基线。我从未把 Cairn 和手机做对照，所以我不能声称它降低了打断 —— 那是意图，不是结果。没有采集任务成功率或任务时长。7 名参与者全是设计背景、在室内完成：他们能告诉我界面读不读得通，不能告诉我它在山上好不好用。
我会先测什么：EMG 手势在低温、戴手套、握登山杖时是否可靠 —— 用真实徒步者、在户外、以手机为基线。
Contact Me
Ready to bring your vision to life? Let's connect and explore the possibilities together. Reach out today!
Work
About
Contact
2025 All rights reserved. Design by Shankun Zhou.
Contact Me
Ready to bring your vision to life? Let's connect and explore the possibilities together. Reach out today!
Work
About
Contact
2025 All rights reserved. Design by Shankun Zhou.
Contact Me
Ready to bring your vision to life? Let's connect and explore the possibilities together. Reach out today!
Work
About
Contact
2025 All rights reserved. Design by Shankun Zhou.